ScaleMatrix Offers Glimpse into San Diego Facility

By Steve Campbell

August 11, 2011

This week Steve Campbell ventured out in San Diego, California to meet the leaders of ScaleMatrix, a startup datacenter company that has chosen to build its behemoth in one of the most unlikely cities possible—at least for anyone with electricity cost concerns in mind.

ScaleMatrix was in the process of building its San Diego datacenter during the on-site interview and throughout the visit, a number of interruptions from the chief builder and others came about, making it clear just how intense a datacenter build can be.

Campbell chatted about datacenter design challenges and benefits of calling San Diego home (despite the clear concerns about power) with the company’s CEO, Mark Ortenzi (pictured), its Vice President, Chris Orlando, and Marc Clark, Director of Business Development.

HPCc: What makes ScaleMatrix different?  It might help to give us a sense of your background and history:

Ortenzi: ScaleMatrix is founded, or based on the cloud platform and partnership with CA Technologies.  I’ve been affiliated with CA and AppLogic and the AppLogic Platform since three years before the CA acquisition. Once CA acquired the AppLogic Platform, I became heavily involved with where they were taking the product and where I wanted to take it based on my knowledge of the product set.  In turn with that, we have (own) a lot of IP in rack design, data center infrastructure and how it works.

When you combine everything that we have; our intellectual property and our rack designs, our cooling models and metrics along with the partnership with CA, cloud computing; you know, there’s a right time and a right place for everything. And I think we found it….

We believe California datacenters needed a new model.  While the traditional 20 or 30-year-old technology that IBM developed back in the day is great, there have to be changes, and why aren’t there being changes made? A lot of the reasons are due to existing infrastructure that customers have and ROI to actually adopt these new platforms.  

With us, building a new data center, a 50,000 square foot data center from scratch, we’re able to take all of these ideas, concepts and actually build them into what we think a data center should really be built as.  When you stack on top of that cloud computing and a Fortune 5000 that has a product now that I can take into SMB or enterprise organizations, along with the co-location piece of the business, there’s really nothing we can’t doto satisfy a customer from the ground up.

So, it’s a true co-location cloud service facility that can take a customer from cradle to grave.

HPCc: I guess I was kind of surprised that you picked San Diego.  I mean, San Diego, to me, is obviously a fabulous place to live, but probably expensive place to do something like this.  So, why San Diego?

Ortenzi:  In the last four or five years, we’ve built out multiple data centers and the acquisition and consolidation of all the data centers in Southern California now have left but two providers in town.  And those two providers are ones outside the county.  One is a large organization publically held where they don’t have that touch and feel that you’re going to get from a customer service based organization like ours.  So, when you look at competition, I don’t think there is anybody in competition with us for what we do and the technology we bring in. It’s a fresh breath of technology that nobody else can offer here in Southern California.

HPCc:  Right, but the cost of electricity here seems to be skyrocketing.  And obviously you use a lot of it here. It is a little bit surprising that the actual raw power it takes to drive a data center is pretty expensive here in Southern California?

Ortenzi:  Yeah, everything’s expensive about it.  Labor is expensive.  Power costs are expensive and infrastructure is expensive.  But we feel based on the technology in Southern California that there’s a lot of companies that need this.  So, when it looks at the co-location piece orthe private cloud infrastructure, the customers want to be close.  They want to have that touch and feel type of aspect, even if we’re managing the infrastructure.  They want to be able to come down and put their finger on it, and there’s a lot of business here in Southern California with all the technology companies, pharmaceutical organizations. There’s not a lack of organizations that need our services here in Southern California.

HPCc:  Okay.  So it’s a point of presence, particularly for co-location that is a big deal here. What’s really unique and really different about your data center?  There a number of data centers here in San Diego.  They’re not publically advertised.  You can’t see them along the street or in the business parks.  They typically have, you know, no name on the building.  But, what’s unique about your offering versus others?

Ortenzi: I’ll start with our concept behind our rack design.  And I’ll let Chris and Marc kind of jump in there.  What’s really unique is our rack design.  So, when you say, you know it’s very expensive, Mark, to have a data center in Southern California.  Electricity is very expensive.  Yeah, average about .14¢ a kilowatt-hour during average run times and up to .28¢ kilowatt-hour based on high peak demand times.  

With my rack design, I’m about 60% more efficient in the way we cool our infrastructure than any other data center in Southern California.  With that said, I’m able to take those cost savings, and that’s what makes a little more sense to have a facility in Southern California.  Our racks are self-contained racks with fire suppression built in.  Shunt trip and we have biometrics on every rack.  Every lock ties back into our control systems that manage everything down to the intake and return for every cabinet that’s monitored on a 3-second basis.  

A customer can actually view into their cabinet and see exactly what temperature the rack is running at for intake and return. They can see the power consumption at the port level for each of their services that we’re running.  We have armed security guards when you walk in the door. This is a very high security lock-down facility.  You’re not going to see that at any other data center in the country. And if you do, I’d love to see it because as far as I know it doesn’t exist.  And we designed the racks ourselves. 

We had them fabricated ourselves. So, everything was built based on our IP and my years of being in the industry and seeing where people are now and why they haven’t transitioned over the last 30 years. They are right back using 30-year datacenter technology, which I think is just absolutely crazy that somebody hasn’t done something like this in the last, you know, five or ten years.  

HPCc:  So, everything about what you’re saying is you’re able to build something brand new, ground up, make use of the latest, and if you like, somewhat future technologies?

Ortenzi:  Absolutely.  Chris, do you want to jump in.

Chris:  Yeah, it’s here because Mark’s put together a pretty fantastic leadership team at the three former presidents of three co-location and data center facilities here running the company.  And so, we’re able to take ideas from all those other companies and dump them onto the table and really start from scratch.  Like you said.

We’re able to leverage today’s latest and greatest technologies to take advantage of all the new efficiencies and design innovations that have happened in the last 10 years, as opposed to all of our competitors in the market who are sitting on 10 year old data center platforms who they’ve either 10, 20 year data center platforms, who they have either forklift upgraded or added on equipment to get to where we are going to be today.  And they’re still going to be falling short as far as efficiency is concerned.

We like to think of ourselves as the last service provider that a customer is ever going to need. Our offering enables us to service people in the traditional co-location model, the hosted model.  All the way through to fully cloud enabled deployments. We do it in a way that enables the customer to switch back and forth between those services at whatever stage in their IP life cycle development they happen to be without contractual penalties.  So, we have a service portability piece of our offering that allows people to start off with, you know, one segment, and grow as their normal progression would happen to be throughout our service platform.  And we think that’s pretty competitive.

HPCc:  Who’s your target customer here in Southern California?  What businesses, types of industries, etc.?

Ortenzi:  Well, we, you know, we’re pretty intimate with the Southern California business landscape.  But I think a healthy business grows with a nice mix of that customer base.  So, we certainly don’t turn away the SMB market.  Those folks who’re the bread butter of a lot of hosting providers.  And a technology getting more and more engrained in every aspect of your business, even those players in the small end of that SMB piece are going to take advantage of our cloud offering.  It’s phenomenal, right now.  Get away from some of thosecapital expenditures and move into our platform makes it very simple and compelling.  

But we also target all the way up through the very, very large enterprise.  We have experience with those customers here in Southern California.  And like you’ve said earlier, San Diego is a great place to live.  You find more and more CEO’s of large companies finding a way to get their headquarters moved here.It’s an interesting thing for us and we’re definitely targeting those companies.

HPCc: The flip side of that, of course, is that San Diego is probably the capital of small businesses. Something like 75% of the businesses employ 100 people or less.

Chris:  Yeah. It’s the largest technology cul de sac in the world.  [Laughter.]

HPCc:  We got SAIC, that’s shrinking and obviously Qualcomm, Waxie, and then you start to fall off the map.  So, having a focus on SMB is probably makes since because, that’s an area that many companiesbelieve they can’t afford to invest in new technology.So pay for usage is clearly a good model to have.    

Chris:  It is.  The SMB is definitely very prevalent here.  But we’ve got the growth and emerging groups that have come up larger too.  You know, the ESETS of the world.  The Intuits.  All those type of folks.  All based here in San Diego.  Or having large operations from a data center, an infrastructure standpoint here.  And those are definitely our targets.

HPCc:  Okay, so it’s kind of like a branch office or remote location?

Chris:  Yep.  

Marc:  Well, and from a cloud computing perspective, think of our audience isn’t just here in Southern California.  Right, it’s really a national audience.  And when you think about cloud computing, our target audience and target customers, from a Cloud computing standpoint, first and foremost, are those folks who don’t want a public cloud.  Ourtag line is ScaleMatrix — private cloud computing provider.  That’s the face of the company first, and we deal with these other things.  But our target really is the companies smalland large that are adoptingthe public cloud computing.  

And whether from a security standpoint; a paranoia standpoint, um, whatever it may be, there’s a lot of folks who want a private solution.  So our main cloud computing solution here is a dedicated solution to that customer.  So that they know where their data is.  They know how it’s secured by the very nature of the AppLogic solution from CA.  There is a level of security there that’s not typically in a public offering.  And then we have solutions that we can layer on top of that for customers that make it even a deeper and more secure solution than that.

HPCc: You mentioned security.  Clearly still a big issue for cloud adoption. What is your approach?You’ve got physical security, no question.  But it’s not the physical security that is the pain for many people in the cloud computing.  But the perception that my data is going to get lost, what about disaster recovery and all those kinds of other issues, you seem to have based your software platform on CA, right?

Ortenzi:  Yes.

HPCc:  So, why CA?  And what’s much better about that from security point of view than other solutions out there?Of which there are many.

Marc:  Well, I’ll let Mark talk a little bit about the why CA and then from a security standpoint, it’s not just that platform by itself, but there are additional layers that can be layered onBusiness continuity, disaster recover. Mark, who founded the company, partially, because of AppLogic, can probably speak to why he chose that.

Ortenzi:  If you look at the market today and who the real players are, you know you’ve got VM out there.  VM as you know is virtualizing in a box.  Being able to manage multiple boxes and you have to add software.  It’s very costly.  So, when you look at theAppLogic platform and the licensing costs and that it’s true cloud computing, you know, being able to take a large chunk of boxes and crate one resource from it and be able to allocate those resources as you see fit and manage from a central location gives you things that nobody else in the market today can give you. Along with that realize that CA has thousands of software products in the market today.  

They have been focused in the enterprise business for a long, long, long time.  What they’re trying to do now is take this enterprise software and move it into the SMB market by putting it within the AppLogic environment and basing it on a licensing model. So, the days you go from a deployment of Wylie, which is a $5,000,000 project in deploying it.  Well, in the SMB market we can’t afford that model.  Now what they’re doing is they’re breaking that down into the SMB market. Well now I can buy it based on a user model and spend $5,000 to acquire the same software that an enterprise spent $5,000,000.

Well, along with that, SiteMinder, one of those security products, has now embedded into the product set.  When we start talking security, well we can give you SiteMinder, we can give you three or four different open source flavors for central authentication and logon.  So, there’s a lot of ways to skin the cat and we have all those based on CA.  CA owns all that software, it’s not hard to take that software and put it into that environment.

HPCc:  Starting from your point of view an advantage is you have a single vendor and you have got everything you need.  You don’t have to go to 10 different vendors and piece it all together?

Chris:  Absolutely not.

HPCc:  That’s a suite. Soup to nuts.

Ortenzi:  Everything I need I have access to through CA.  CA is on track to do this and that we’re in the process of doing it.  When we talk SiteMinder, great, that’s one of thousands of products they own.  Let’s talk Wiley, it’s already wrapped up and put into the environment.  Let’s talk KJO or Clarity, wrapped up and put into AppLogic, andCirrus 9 just did Albacore.  

There’s a lot of enterprise applications now that reside within the AppLogic virtualization layer that allows SMB’s to buy it based on a licensing model.  I don’t know anybody else in the world that can do that.

HPCc:  If in a co-location model a little different, a customer comes in and is not based on CA. Is that a problem?

Chris: Not at all.  No.  You know, we’ve got, one of the things that is so different about ScaleMatrix is that, you know, we’re a cloud Solutions provider, but we’re, we’re based around, you know, very, very well built out infrastructure.  So that, that infrastructure platform provides a launch point for lots of different services.  It means that we’re able to provide fantastic backing for that cloud offering, but co-location becomes a natural by-line of that development. A class 3 center built up with all the accouterments that you can expect that’s even if a customer isn’t tied into the CA platform, yeah, we like to start them out there.  And then as their needs grow and they expand outside of the standard traditional co-location market, which almost everybody’s looking to cloud in some way today. 

Even if they don’t know what their requirements are, they’re trying to figure it out.  We like to help them along in that path and develop them so that they can create a hybrid solution.  All those traditional co-location services can be hosted here, and if they grow towards the cloud solution and the CA Platform, were here to help.

HPCc: Now, presuming its all X-86 based architecture?

Chris:  For the CA platform. Yes.

HPCc:  Right.  So doesn’t much matter whether its Intel or AMD, but I image its Intel.

Ortenzi:  We have partnered directly with Intel in the past and we are now a platinum partner.  So we do like to work directly with Intel for a lot of reasons: Good road mapping, access to product set in advance so we can test to it.  We get some market recognition with them and we have good relationships over the last 10 years.

Chris:  I think that’s one of our differentiators, is the level of strategic partnering that we’vedone.  The company was founded by Mark, and in part because of his relationship with CA.  But even beyond that and outside that scope, we have long standing relationships with Intel and Brocade.  And, a number of other partners that are really key to the operation of this business.  And without those strong relations, we would not be able to pick up the phone, get senior level leadership to throw resources at problems that we’ve got.  I don’t know if anybody else is going to be able to step in like that in those particular types of situations.

HPCc:  What’s the storage architecture? This is a big piece, where does the data actually go?

Ortenzi:  Yeah.  Where, where does the data go?  So, based on the architecture of AppLogic and the IP mesh that it runs on. The storage actually runs within the X86 technology.  However beefy you want the boxes, however much storage you want within a box environment, it’salways running in an HA factor with multiple controllers within the environment.  So, if you’re running a specific application here within the AppLogic grid, it’s also running on a different piece over here in case this drivefails.Pretend it’s like a RAID 5 configuration within an X86 platform of hundredsof machines.  When we talk HA, I could walk up pull a server out of that cluster of servers, the controller will go, “Uh, oh.  We have a problem, re-direct, it’ll point over towards where it’s running and you’re back up and running.

HPCc:  So, you’re got mirroring capability?

Ortenzi:  Across the platform.  Right.

Chris:  So with an AppLogic, youcan actually create data stores and data silos across the combined amount of storage in the grid.

Marc:  One of the other things that make us different is that we are nimble.

We went out to the market and heard that the target customer out there that the other companies aren’t taking care of their technology the way they wanted.  So, being nimble and being a small company, we’re able to go, “Hey. There’s aneed.  Let’s go fill it.”  As director of Business Development, some of the things I look for when I’m talking to customers is what do they really need?  What do people want within the cloud?  We’re not going to do public cloud. Now, guess what? 

There are customers who want burst-ability.  Right.  What does that mean?  Well, can we potentially put up a product that allows a private cloud customer 10 days a year or two weeks out of the year when they have these spikes burst over into a public area. That’s a shared area that they’re okay with though for that little piece of time.  That’s some of the types of things that Amazon might do, for example.  

I saw that out in the market.  Where I saw that Amazon is doing that for folks and letting them pop up into their atmosphere for a little bit.  And then coming back down to a smaller private solution.So, those are the types of things that we can do and look at being a smaller company but being nimble.  But also having the infrastructure and the know-how within the executive team to put together those types of products within a cloud environment.

HPCc: Nimbleness and flexibility that’s certainly of good.  What I really like is that you’re actually going to customers and asking them what they need. You’ve actually gone to identify what the problems are and then you’re building a solution to apply to the problem.  

Marc:  Right.  The company was founded because Mark and the other partners saw a problem in the marketplace that needed to be fulfilled.  private cloud Solutions.  Cloud Solutions that are now private.  And not private from the standpoint of what Dell and EMC and these folks want you to do, and that’s just buy their hardware and put it onyour premise.

Bring us you problems and let us become your infrastructure as a service providerand let us take care of you.  The company was founded on a problem out there in the marketplace, an opportunity.  As we see more and more opportunities within the community we’re able to adapt and potentially add products where it makes sense for us and for our customers.

HPCc:  We touched briefly on the security and data.  What about disaster recovery?  What is a plan?  If you’re on there, you’re working.  Boom.  Big failure.  What’s the customer going to see?

Ortenzi:  Well, there are several ways to look at that.  We give offerings to the customers.  So the days of back up I think are moving away.  But we still do give back up, because I think it is pertinent that you also have your data backed up.  As anybody who’s lost data before can tell you that.  Or its been mandated.  The second part, I really think things are going to business continuity.  True fail-over.  Right?  You got a 7-office location – 10, 20, whatever it might be, and you have a problem.  Whether it’s the data center itself. The hardware infrastructure.  Internet connectivity.  Whatever it might be.  You can’t stop working; the cost in the downtime of what the company has to take on when they have an issue is dramatic.  

The whole idea with business continuity in the AppLogic platform is true fail-over.  Remember, we have 12 data centers across the country.  So, we have that offering and we can help the customers do true business continuity.  If you do have an issue, you’re really not going to know it, the your latency might be a little bit bigger because you’re going to go across the country.  And, of course, we give you the back-up option.  And you could pay for back up and we’ll push off to a remote location and it’s all private within your infrastructure or you’ll be part of the shared platform for data.  We give the customers all these options.  It’s what do they want to take advantage off.  I can lead a horse to water but I can’t make him drink’

HPCc:  So, you mention 12 data centers across the country.  So this is not the only data center?

Ortenzi:  We partnered with Forsythe, which has 11 other data centers around the country.  And we’ll also be going into Europe and the Asia market as soon as this datacenter is complete.  So we do have relationships.  They do want us to move into those territories based on our cloud Platform and a couple of other things we’re doing with them.  I’m just not ready to do it yet.

HPCc:  One of the areasthat’s always difficult to quantifyin the pay-for-usage model,which cloud computing promises versus have – owning it yourself. How do you measure the economics and how do you quantify the savings?

Marc:  There are different areas, obviously.  I was just recently at a CIO roundtable where several San Diego CIO’s were talking about their need to upgrade their hardware.  But with the current economy they’re just not getting the budgets approved. We can go to those CIO’s and say, “Have you considered an OPeX solution as opposed to a CAPeX?” CIOs can get three or four thousand a month for a private cloud approved as an OPeX. But I can’t get the $50,000 or $100,000 or $200,000 for the hardware. In addition to saving the CAPeX we can also provide better security and save on cost of utilities.

So that’s really the model.  It’s also about what’s your current environment.  What are you doing today?  Is it time to move into some additional hardware solutions? We don’t have the budget. Come here to ScaleMatrix and let us, you know, help you pencil out an OPeX opportunity that is flexible.  

So, we’re going to have some customers who are going to come in here and they’re just going to take their equipment to day and they’re going to Colo.  And in a year or two from now they’re going to say, “You know what? I’m ready to move into the cloud.”  And to Chris’ point earlier, we have those options.  And they’ll be ready to move into a private cloud.  And then as we potentially offer maybe some sort of public cloud as well.
When you talk about penciling it out we do an ROI for every single customer. What are you doing today?  What’s your current environment? 

Also, the AppLogic platform allows us to really be getting more efficient.  I had a customer who had a traditional 27-server environment that he was looking to deploy.  When we looked at what he was actually putting on the servers in the new environment, even in the VMWare environment, and we were able to reduce his need for servers here almost by half. So, not only does he have an OPeX now, but that OPeX is based on a much smaller number that we had to go out and buy, and so his overall costs have been a lot more efficient.

HPCc:  Basically, you can clearly ride Moore’s Law and processor performance better than a customer.  As processors double in speed every 18 months to two years, they can get the similar amount of work through half the actual widgets. To them it represents a consolidation solution.

Marc: Our private cloud offering is not going to compare to a public cloud.  We’re not going to compare to an Amazon or a Rackspace who’s offering at pennies for an hour.  Because again, they have thousands of customers on that.  But our offering is really for those customers who go, “Oh, wow!  Amazon went down again last month for two days.  Or, Oh, Google got hacked again.”  And they go, “Oh, I just can’t.  I was going to.  I thought I was convinced.  It’s not a red herring.  It’s not a falsehood.  It’s true that there are concerns about security and those customers who have those concerns want a private solution.  And from that perspective we are much more cost efficient than somebody doing all this on their own.

HPCc: Is there any kind of minimum size to get started?

Marc: Yeah.  About four servers depending on the size of the servers.Anywhere from about 2500 hundred on a monthlyfor a four server solution.  That’s going to give you about 32 cores, 16 gig of RAM per server, times four, at least a terabyte perserver for four terabytes.  So that’s sort of minimum.  The reason we do that minimum is really to protect the customer.  Because even if they have less need than that, we want them to make that they have a high ability.

HPCc:  Any last points you want to make?

Marc:  Yeah. Traditional clusters of servers and how it’s built, traditionally we’d go out and buy those 20 servers. We build load balancers, web servers, data base servers, media servers, maybe some exchange servers, Firewalls, etc.  First of all, we got to order the equipment, which is very expensive.  Then how long does it take to really rack and stack, load all the software, configure these boxes and get them up and running.  Now, if you’re a real good team, you can do it in a week.  If you’re corporate, you’relucky if you got it done in a month to three months.

In the AppLogic platform we build this infrastructure and put it up within days for you.When you are all built with this configuration of in traditional load balancers and web servers, etc. your utilization is about 25 to 30% within that infrastructure.  You can’t go and stick something else on top of that infrastructure based on what the platform might be.  With AppLogic, all it is is spinning it up another instance of this to bring your userability from a 30% all the way up into the 70% range, still running in an HA factor.  

Now, if you’ve created it once, you never have to create it again in AppLogic.  If I built this whole web servers and, and load balancer configuration and I have another customer that calls up and says, “Hey.  I need the same set-up.”  Well, it’s a drag and drop.

The other big thing which we see coming down the road, and I’m real excited about, is with that SMB market, whether it’s a start-up or a traditional company is the business in a box.  If you can come to us, we can deploy all your VDI solution,email, or phone, CIP, DID, all in the cloud.  So, as a business, why would you go out and spend $30,000 or $60,000 all on a phone system? I can do it within the cloud within probably about 30 minutes, all CIP, DID, cell phone or desk phone based. I can give you all your Microsoft, Office Exchange, Share Point, based on our SQL licensing agreement. 

We can spin a whole company up overnight with such minimal costs versus anybody trying to do it themselves.  Imagine what you’re going to pay an exchange engineer. 

Imagine what you’re going to pay a Share Point engineer, aVDI solution, the Microsoft Office guys, the server guys, the power guys and the HVAC guys.  You can do this through a portal that we’re creating in partnership with CA, which will be cloud Commons, that you can actually deploy this within my environment on the fly and start your whole business up overnight, with no CAPeX.

HPCc: That’s an attractive proposition, no question.

Ortenzi:  I wish I’d have had this when I started up my businesses, right? We are the platform that’s doing this.  So it’s really not offered in this type of level, right?  And thatwould all be in the cloud.

HPCc:  So, do you eat your own dog food?

Ortenzi:  Absolutely!  Absolutely.  Everything we do is all in the cloud.  It’s all in the cloud. And if it’s not in the cloud, we’re putting it in the cloud because it hasn’t been. I’m going to make up a new word, cloudified!  Right.  So, we take it and put it in that environment.  So everything that we’re doing, and if it’s not in their now, it’s being put in as we speak.  

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion XL — were added to the benchmark suite as MLPerf continues Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing power it brings to artificial intelligence.  Nvidia's DGX Read more…

Call for Participation in Workshop on Potential NSF CISE Quantum Initiative

March 26, 2024

Editor’s Note: Next month there will be a workshop to discuss what a quantum initiative led by NSF’s Computer, Information Science and Engineering (CISE) directorate could entail. The details are posted below in a Ca Read more…

Waseda U. Researchers Reports New Quantum Algorithm for Speeding Optimization

March 25, 2024

Optimization problems cover a wide range of applications and are often cited as good candidates for quantum computing. However, the execution time for constrained combinatorial optimization applications on quantum device Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at the network layer threatens to make bigger and brawnier pro Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HBM3E memory as well as the the ability to train 1 trillion pa Read more…

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HB Read more…

Nvidia Looks to Accelerate GenAI Adoption with NIM

March 19, 2024

Today at the GPU Technology Conference, Nvidia launched a new offering aimed at helping customers quickly deploy their generative AI applications in a secure, s Read more…

The Generative AI Future Is Now, Nvidia’s Huang Says

March 19, 2024

We are in the early days of a transformative shift in how business gets done thanks to the advent of generative AI, according to Nvidia CEO and cofounder Jensen Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Nvidia Showcases Quantum Cloud, Expanding Quantum Portfolio at GTC24

March 18, 2024

Nvidia’s barrage of quantum news at GTC24 this week includes new products, signature collaborations, and a new Nvidia Quantum Cloud for quantum developers. Wh Read more…

Alibaba Shuts Down its Quantum Computing Effort

November 30, 2023

In case you missed it, China’s e-commerce giant Alibaba has shut down its quantum computing research effort. It’s not entirely clear what drove the change. Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Leading Solution Providers

Contributors

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

Google Introduces ‘Hypercomputer’ to Its AI Infrastructure

December 11, 2023

Google ran out of monikers to describe its new AI system released on December 7. Supercomputer perhaps wasn't an apt description, so it settled on Hypercomputer Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Intel Won’t Have a Xeon Max Chip with New Emerald Rapids CPU

December 14, 2023

As expected, Intel officially announced its 5th generation Xeon server chips codenamed Emerald Rapids at an event in New York City, where the focus was really o Read more…

IBM Quantum Summit: Two New QPUs, Upgraded Qiskit, 10-year Roadmap and More

December 4, 2023

IBM kicks off its annual Quantum Summit today and will announce a broad range of advances including its much-anticipated 1121-qubit Condor QPU, a smaller 133-qu Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire